Local anaesthetic sympathetic blockade for complex regional pain syndrome

Transcription

1 The University of Notre Dame Australia Physiotherapy Papers and Journal Articles School of Physiotherapy 2016 Local anaesthetic sympathetic blockade for complex regional pain syndrome N O'Connell B Wand The University of Notre Dame Australia, benedict.wand@nd.edu.au W Gibson University of Notre Dame Australia, william.gibson@nd.edu.au D Carr F Birklein See next page for additional authors Follow this and additional works at: Part of the Rehabilitation and Therapy Commons This article was originally published as: O'Connell, N., Wand, B., Gibson, W., Carr, D., Birklein, F., & Stanton, T. (2016). Local anaesthetic sympathetic blockade for complex regional pain syndrome. Cochrane Database of Systematic Reviews, 2016 (7). Original article available here: This article is posted on ResearchOnline@ND at For more information, please contact researchonline@nd.edu.au.

2 Authors N O'Connell, B Wand, W Gibson, D Carr, F Birklein, and T Stanton This article is available at ResearchOnline@ND:

3 This is the published version of the following review: O Connell, N., Wand, B., Gibson, W., Carr, D., Birklein, F., and Stanton, T. (2016). Local anaesthetic sympathetic blockade for complex regional pain syndrome (Review). Cochrane Database of Systematic Reviews, 2016(7). doi: / CD pub4. Which has been published in final form at This article may be used for non-commercial purposes in accordance with Wiley Terms and Conditions for self-archiving.

4 Cochrane Database of Systematic Reviews Local anaesthetic sympathetic blockade for complex regional pain syndrome(review) O ConnellNE,WandBM,GibsonW,CarrDB,BirkleinF,StantonTR O ConnellNE,WandBM,GibsonW,CarrDB,BirkleinF,StantonTR. Local anaesthetic sympathetic blockade for complex regional pain syndrome. Cochrane Database of Systematic Reviews 2016, Issue 7. Art. No.: CD DOI: / CD pub4. Local anaesthetic sympathetic blockade for complex regional pain syndrome(review) Copyright 2016 The Cochrane Collaboration. Published by John Wiley& Sons, Ltd.

5 T A B L E O F C O N T E N T S HEADER ABSTRACT PLAIN LANGUAGE SUMMARY SUMMARY OF FINDINGS FOR THE MAIN COMPARISON BACKGROUND OBJECTIVES METHODS RESULTS Figure Figure Figure DISCUSSION AUTHORS CONCLUSIONS ACKNOWLEDGEMENTS REFERENCES CHARACTERISTICS OF STUDIES DATA AND ANALYSES ADDITIONAL TABLES APPENDICES WHAT S NEW HISTORY CONTRIBUTIONS OF AUTHORS DECLARATIONS OF INTEREST SOURCES OF SUPPORT DIFFERENCES BETWEEN PROTOCOL AND REVIEW NOTES INDEX TERMS i

6 [Intervention Review] Local anaesthetic sympathetic blockade for complex regional pain syndrome Neil E O Connell 1, Benedict M Wand 2, William Gibson 2, Daniel B Carr 3, Frank Birklein 4, Tasha R Stanton 5,6 1 Department of Clinical Sciences/Health Economics Research Group, Institute of Environment, Health and Societies, Brunel University, Uxbridge, UK. 2 School of Physiotherapy, The University of Notre Dame Australia, Fremantle, Australia. 3 Pain Research, Education and Policy (PREP) Program, Department of Public Health and Community Medicine, Tufts University School of Medicine, Boston, Massachusetts, USA. 4 University Medical Centre, Johannes Gutenberg University, Mainz, Germany. 5 Neuroscience Research Australia, Randwick, Australia. 6 The Sansom Institute for Health Research, School of Health Sciences, University of South Australia, Adelaide, Australia Contact address: Tasha R Stanton, The Sansom Institute for Health Research, School of Health Sciences, University of South Australia, GPO Box 2471, Adelaide, South Australia, 5001, Australia. t.stanton@neura.edu.au. tasha.stanton@unisa.edu.au. Editorial group: Cochrane Pain, Palliative and Supportive Care Group. Publication status and date: Stable (no update expected for reasons given in What s new ), published in Issue 7, Review content assessed as up-to-date: 16 September Citation: O Connell NE, Wand BM, Gibson W, Carr DB, Birklein F, Stanton TR. Local anaesthetic sympathetic blockade for complex regional pain syndrome. Cochrane Database of Systematic Reviews 2016, Issue 7. Art. No.: CD DOI: / CD pub4. Background A B S T R A C T This review is an update of a previously published review in the Cochrane Database of Systematic Reviews, 2005, Issue 4 (and last updated in the Cochrane Database of Systematic Reviews, 2013 issue 8), on local anaesthetic blockade (LASB) of the sympathetic chain to treat people with complex regional pain syndrome (CRPS). Objectives To assess the efficacy of LASB for the treatment of pain in CRPS and to evaluate the incidence of adverse effects of the procedure. Search methods For this update we searched the Cochrane Central Register of Controlled Trials (CENTRAL) (2015, Issue 9), MEDLINE (Ovid), EMBASE (Ovid), LILACS (Birme), conference abstracts of the World Congresses of the International Association for the Study of Pain, and various clinical trial registers up to September We also searched bibliographies from retrieved articles for additional studies. Selection criteria We considered randomised controlled trials (RCTs) that evaluated the effect of sympathetic blockade with local anaesthetics in children or adults with CRPS compared to placebo, no treatment, or alternative treatments. Data collection and analysis We used standard methodological procedures expected by Cochrane. The outcomes of interest were reduction in pain intensity, the proportion who achieved moderate or substantial pain relief, the duration of pain relief, and the presence of adverse effects in each treatment arm. We assessed the evidence using GRADE (Grading of Recommendations Assessment, Development and Evaluation) and created a Summary of findings table. 1

7 Main results We included an additional four studies (N = 154) in this update. For this update, we excluded studies that did not follow up patients for more than 48 hours. As a result, we excluded four studies from the previous review in this update. Overall we included 12 studies (N = 461), all of which we judged to be at high or unclear risk of bias. Overall, the quality of evidence was low to very low, downgraded due to limitations, inconsistency, imprecision, indirectness, or a combination of these. Two small studies compared LASB to placebo/sham (N = 32). They did not demonstrate significant short-term benefit for LASB for pain intensity (moderate quality evidence). One small study (N = 36) at high risk of bias compared thoracic sympathetic block with corticosteroid and local anaesthetic versus injection of the same agents into the subcutaneous space, reporting statistically significant and clinically important differences in pain intensity at one-year follow-up but not at short term follow-up (very low quality evidence). Of two studies that investigated LASB as an addition to rehabilitation treatment, the only study that reported pain outcomes demonstrated no additional benefit from LASB (very low quality evidence). Eight small randomised studies compared sympathetic blockade to various other active interventions. Most studies found no difference in pain outcomes between sympathetic block versus other active treatments (low to very low quality evidence). One small study compared ultrasound-guided LASB with non-guided LASB and found no clinically important difference in pain outcomes (very low quality evidence). Six studies reported adverse events, all with minor effects reported. Authors conclusions This update s results are similar to the previous versions of this systematic review, and the main conclusions are unchanged. There remains a scarcity of published evidence and a lack of high quality evidence to support or refute the use of local anaesthetic sympathetic blockade for CRPS. From the existing evidence, it is not possible to draw firm conclusions regarding the efficacy or safety of this intervention, but the limited data available do not suggest that LASB is effective for reducing pain in CRPS. P L A I N L A N G U A G E S U M M A R Y Local anaesthetic sympathetic blockade for complex regional pain syndrome Background Local anaesthetic sympathetic blockade (LASB) is a common treatment for complex regional pain syndrome (CRPS). It involves blocking the activity of sympathetic nerves alongside the spine. The sympathetic nervous system mainly controls unconscious actions such as heart rate, blood flow, and perspiration. The injection of a local anaesthetic drug around the nerves temporarily blocks the function of the nerves. This updated review aimed to summarise the available evidence regarding whether LASB is effective at reducing pain in CRPS, how long any pain relief might last, and whether LASB is safe. Key results and quality of the evidence In September 2015, we found a limited number of small trials, all of which had design flaws. We did not find evidence that LASB was better than placebo in reducing pain, or that it provided additional pain relief when added to rehabilitation. While a number of small studies compared LASB to other treatments, most did not find that LASB was better. One small study found that injecting the thoracic (upper back) sympathetic nerves with local anaesthetic and steroid was better than injecting the same drugs just under the skin at oneyear follow-up, but the study may have been prone to bias. Only six studies reported on the type and amount of side effects. These studies reported only minor side effects, but since some studies did not report this information we can draw no firm conclusions about the safety of LASB. The evidence was mostly of low or very low quality. Overall, the evidence is limited, conflicting, and of low quality. While we cannot draw strong conclusions, the existing evidence is not encouraging. 2

8 S U M M A R Y O F F I N D I N G S F O R T H E M A I N C O M P A R I S O N [Explanation] Patient or population: adults with CRPS Setting: secondary care Intervention/ comparison: LASB vs various comparisons Outcome: pain intensity 0-10 (VAS or NRS) Comparison Studies No of participants (studies) Result (effect estimates reported where available from study report) Quality of the evidence (GRADE) LASB vs placebo Aydemir 2006; Price (2) No significant betweengroup difference M oderate a Thoracic LASB + steroid vs subcutaneous local anaesthetic+ steroid Rocha (1) Favours LASB Mean difference (0-10 scale) One month 1.25 (95% CI 3.2 to 0.7) One year 2.39 (95%CI 4.72 to 0.06) Very low b LASB vs ultrasound block LASB vs IVRB guanethidine LASB lumbar plexus vs pulsed radiofrequency lumbar plexus LASB (lidocaine + clonidine) vs IVRB (lidocaine + clonidine) Aydemir (1) No significant betweengroup difference Bonelli (1) No significant betweengroup difference Freitas (1) No significant betweengroup difference Nascimento (1) No significant betweengroup difference Low c Very low b Very low b Very low b LASB + PT+ pharmacological vs PT + pharmacological Rodriguez (1) Favours SGB group Very low b LASB + PT vs PT Zeng (1) No significant betweengroup difference Very low b Continuous LASB vs continuous brachial plexus block Toshniwal (1) Favours brachial plexus block Low c 3

9 Image-guided LASB vs nonimage-guided LASB Yoo (1) Mean difference 2 weeks postinjection 0.58 (95%CI 1.51 to 0.35) 4 weeks postinjection 0.74 (95%CI 1.36 to 0.12) Very low d Outcome: hand pain 0-3 scale LASB vs oral corticosteroids Lim (1) 15 day follow-up, no significant between-group difference at 30 day follow-up 0.4 (95% CI 0.69 to 0. 11), favours LASB with steroid Very low d Outcome: duration of pain relief LASB bupivacaine + BTA vs LASB bupivacaine Carroll (1) Increased duration of relief with BTA Median time to analgesic failure (days): LASB bupivacaine + BTA 71 (95% CI 12 to 253) LASB bupivacaine 10 (95%CI 0 to 12) Low c a Downgraded once for imprecision. b Downgraded three times for limitations, inconsistency, and imprecision. c Downgraded twice for inconsistency and imprecision. d Downgraded four times for limitations, inconsistency, indirectness, and imprecision. B A C K G R O U N D This review is an update of a previously published review in the Cochrane Database of Systematic Reviews, 2005, Issue 4 (and last updated in the Cochrane Database of Systematic Reviews, 2013 Issue 8), on local anaesthetic sympathetic blockade for complex regional pain syndrome. Description of the condition Complex regional pain syndrome (CRPS) is an umbrella term for a variety of clinical presentations characterised by chronic persistent pain that is disproportionate to any preceding injury (if any) and that is not restricted anatomically to the distribution of a specific peripheral nerve (Bruehl 2010). The International Association for the Study of Pain (IASP) introduced the diagnostic label of CRPS in the 1990s (Merskey 1994), and since then, others have updated it in an attempt to improve its specificity (Harden 2006; Harden 2010). We present these modified diagnostic criteria (the Budapest criteria ) in Table 1. The term CRPS encompasses a variety of earlier diagnostic terms, including reflex sympathetic dys- 4

10 trophy (RSD), reflex neurovascular dystrophy, Sudeck s atrophy, causalgia, and algodystrophy/algoneurodystrophy (Stanton-Hicks 1995). CRPS can be classified into two subtypes: CRPS-I, in which there is no identified peripheral nerve injury, and CRPS-II, where symptoms are associated with a definable nerve lesion (Harden 2006). This distinction is not always easily made (Harden 2006). Both subtypes of CRPS are characterised by severe pain that is disproportionate to the inciting event, most commonly affecting the hand or foot but sometimes spreading to other body regions (Stanton-Hicks 2002; Van Rijn 2011). Additionally CRPS presents with some or all of the following symptoms in the affected body parts: sensory disturbances; temperature changes; abnormal patterns of perspiration; swelling/oedema; reduced joint range of motion; movement abnormalities such as weakness, tremor, or dystonia; trophic changes such as skin atrophy, altered hair and nail growth, or localised osteoporotic changes (Bruehl 2010; De Mos 2009; Shipton 2009); and alterations in body perception (Lewis 2007; Lotze 2007; Moseley 2006). CRPS occurs most commonly following wrist fracture and subsequent immobilisation. However, cases can potentially occur after relatively minor trauma and may even occur spontaneously, albeit rarely (De Mos 2007; De Mos 2008; Sandroni 2003). The underlying pathophysiological mechanisms of CRPS are incompletely understood, although there is growing consensus that it is primarily a disorder of the nervous system. Research has identified abnormalities in the tissues of the affected area and the peripheral and central nervous systems (Jänig 2003; Marinus 2011). These include signs of increased neurogenic inflammation (Birklein 2001; Schinkel 2006; Schmelz 2001), an altered local immune response (Birklein 2014; Tan 2005), altered activity in the sympathetic nervous system (SNS) (Drummond 2004; Niehof 2006), increased sensitivity to normal SNS activity (Albrecht 2006; Ali 2000; Drummond 2001), and local tissue hypoxia (Birklein 2000; Koban 2003). Studies have also demonstrated changes in the brain in CRPS (Swart 2009), including alterations of the cortical (higher brain) representation of the affected body part (Maihöfner 2004; Pleger 2006), localised reductions in grey matter density and connectivity (Geha 2008), and altered inhibitory control (Schwenkreis 2003). Description of the intervention Sympathetic blockade includes procedures that aim to temporarily impede the local function of the sympathetic nervous system. Usually an anaesthesiologist performs the procedure, injecting local anaesthetic directly into sympathetic neural structures that serve the affected limb(s) such as the stellate ganglion or the lumbar sympathetic chain (Nelson 2006). Radiologic guidance such as fluoroscopy or computerised tomography (CT) scan often ensures the accuracy of needle tip placement, and successful blockade is often monitored by direct (e.g., galvanic skin response) or indirect (increase in blood flow to the extremity or increase in temperature) assessment (Breivik 2009). This approach is distinct from the injection of neurolytic agents in an effort to destroy sympathetic nerves. LASBs are also commonly called stellate ganglion blockades (SGB) or, when performed in the lower body, lumbar sympathetic blockades (LSB). How the intervention might work People with persistent pain following nerve injury have long been observed to have abnormalities of autonomic nervous system function in the affected limb (temperature, blood flow, sweating) and abnormal skin texture or hair and nail growth attributed, at least in part, to local autonomic dysfunction (Bruehl 2010; De Mos 2009). Early uncontrolled observations of persistent improvement in signs and symptoms following local anaesthetic sympathetic blockade in people with what is now termed CRPS suggested that excessive sympathetic activity provoked or perpetuated this type of persistent pain (Campbell 1996). However, recent evidence regarding adrenaline content in venous effluents from affected limbs has not supported this hypothesis and suggests instead that any benefit of sympathetic blockade in CRPS may reflect transient reversal of a heightened local sensitivity to adrenaline (Binder 2009). These clinical impressions of persistent benefit from transient local anaesthetic sympathetic blockade in CRPS, reinforced by similar longstanding impressions of prolonged benefit after temporary local anaesthetics blockade in peripheral neuralgias, led to the incorporation of sympathetic block into current consensus treatment algorithms for CRPS (Carr 2011), although doubt remains over the contribution of the sympathetic nervous system to pain and the concept of sympathetically maintained pain in CRPS (Harden 2013). Why it is important to do this review Despite preclinical evidence that suggests the sympathetic nervous system is involved in the pathophysiology of CRPS, there is debate surrounding the contribution of the sympathetic nervous system to the clinical syndrome (Ochoa 1995; Schott 1995; Verdugo 1994a; Verdugo 1994b). The value of blocking the sympathetic nervous system is also disputed (Fine 1994; Hogan 1997; Jadad 1995; Verdugo 1994a). It is therefore important to evaluate the efficacy of sympathetic blockade with local anaesthetic in the treatment of CRPS. A meta-analysis of the effect of sympathetic blockade with local anaesthetics in people with CRPS reported that up to 44% of those subjected to sympathetic blockade would be expected to have no pain relief. Due to the lack of randomised controlled trials, investigators obtained this estimate from pooling the results of observational studies (Cepeda 2002). Moreover, the review only evaluated English-language studies, and it could have overlooked relevant RCTs. Hence, to overcome this limitation, we decided to perform a systematic review of the literature with no language restriction to determine both the efficacy and safety of 5

11 sympathetic blockade with local anaesthetics to alleviate pain in people with CRPS. into the stellate ganglion, because this procedure does not block sympathetic activity. O B J E C T I V E S To assess the efficacy of LASB for the treatment of pain in CRPS and to evaluate the incidence of adverse effects of the procedure. M E T H O D S Criteria for considering studies for this review Types of studies We considered randomised controlled trials (RCTs). As blinding of sympathetic block is not always possible, we included trials that were either double-blind, single-blind, or open. We included studies that compared LASB with placebo interventions, no treatment, or alternative interventions. We also included studies that investigated the effect of adding LASB to other interventions. Types of participants We included studies that evaluated the effect of sympathetic blockade with local anaesthetics to treat CRPS in children or adults. We included studies even if the authors did not describe the constellation of symptoms necessary to diagnose CRPS and stated only that patients with RSD/CRPS were included. We took this approach to avoid excluding any of the relatively few RCTs of this intervention. We placed no restrictions regarding the number of participants recruited to trials. We excluded trials that evaluated sympathetic blockade for other pain syndromes such as radiculopathy, herpes zoster, postherpetic neuralgia, fibromyalgia, or phantom pain. Types of interventions We included studies that evaluated selective sympathetic blockade with local anaesthetics. We excluded studies that only evaluated somatic nerve blocks or studies that evaluated the effect of local anaesthetics or sympatholytic drugs administered orally, intravenously, or epidurally. We excluded studies that reported the results of combined sympatholytic therapies, such as surgical sympathectomy or guanethidine intravenous regional block plus local anaesthetic blockade of the sympathetic chain. We also excluded studies of ganglionide local opioid analgesia (GLOA), a technique in which clinicians locally inject opioids such as buprenorphine Types of outcome measures The outcomes of interest were pain intensity levels, duration of pain relief. and adverse events. For this update, we excluded studies that had only immediate follow-up data ( 48 h), because this information provides little clinically relevant information about the effectiveness of this treatment. We applied this new criterion to studies that had been included in previous updates of this review. Search methods for identification of studies For this update, we used identical search strategies to that of our 2013 review update. For the search strategies, see Appendix 1 for the Cochrane Central Register of Controlled Trials (CEN- TRAL), Appendix 2 for MEDLINE, Appendix 3 for EMBASE, and Appendix 4 for LILACS. We performed the search for the original review from November 2003 to January 2004 updated it on 17 November 2011 and 22 November 2012 (2013 update). The present update encompasses searches run from 22 November 2012 to 16 September We evaluated non-english language papers for inclusion. For the 2016 update, we did not search the Cochrane Pain, Palliative and Supportive Care Group Specialised Register, as it is no longer updated. Electronic searches We searched the following databases for the update of this review. CENTRAL (The Cochrane Library 2015, Issue 9). MEDLINE (Ovid) (1966 to September 2015). EMBASE (Ovid) (1974 to September 2015). LILACS (Birme) (1982 to September 2015). Searching other resources Reference lists We searched the bibliographies of retrieved articles for additional studies. Unpublished studies In order to minimise the impact of publication bias, we reviewed conference abstracts of the World Congresses of the International Association for the Study of Pain from 1995 up to For this update, we expanded the search of the original review by also searching relevant clinical trial registers (from inception) for upcoming trials. We searched the following clinical trial registers: 6

12 the controlled trials register (15 October 2015; the United States National Institute of Health service ClinicalTrials.gov (15 October 2015; the Australian New Zealand Clinical trials register (15 October 2015; and the European Clinical Trials Register (7 December 2012; Personal contact We attempted to communicate with authors if we needed additional information that was not provided in the trial report. In addition, we provided the reference list of included studies to experts in the field to determine if any additional references were appropriate for the review. 8. Treatment characteristics: site of sympathetic block (cervical or lumbar), type of local anaesthetic used (including concentration and volume), evaluation of the technical adequacy of the block, duration of follow-up, duration of the pain relief, number of blocks performed, method of pain assessment, and presence of complications or adverse effects. 9. Information on postprocedure analgesic requirements. 10. Information on conflicts of interest and statements of study support. If authors reported pain intensity using a visual analogue scale or numeric rating scale, we extracted the mean and standard deviation of pain intensity in each study arm. If authors reported pain relief, we extracted the proportion of participants in each category of pain relief. Data collection and analysis Selection of studies Two review authors independently read each of the titles and abstracts of the reports identified by the search and discarded narrative reviews, case series, and case reports. If there was no abstract, we retrieved the full-text report. If there was disagreement, the authors met to reach consensus, consulting an independent third review author if necessary. We retrieved in full all abstracts and reports that made reference to a trial of sympathetic blockade with local anaesthetics. Two review authors then independently assessed the full-text articles. We did not anonymise the reports for the assessment. Data extraction and management Two review authors independently extracted the data. If there was disagreement, they met to reach consensus, consulting an independent third review author if necessary. We extracted the following data from each study. 1. Study details: study design (parallel or cross-over), method of randomisation, presence or absence of blinding. 2. Demographic characteristics: age, sex, number of participants recruited, number of study withdrawals or dropouts, if any. 3. Participant clinical characteristics: duration of pain before sympathetic block, site of pain (arm, leg, mixed, or other such as facial). 4. Type of noxious initiating event (if known): surgery, fracture, crush injury, projectile, or stab injury. 5. Type of tissue injured: nerve, soft tissue, bone. 6. Presence of medico-legal factors that may influence the experience of pain and the outcomes of therapeutic interventions. 7. Concomitant treatments that may affect outcome: antidepressants, physical therapy, etc. Assessment of risk of bias in included studies We used a modified version of the Cochrane Risk of bias tool with additional domains added in response to the recommendations of Moore On this basis we added two domains, size and duration, using the thresholds for judgement suggested by Moore We have not added the outcome domain as this is covered already by our choice of primary outcome measures. Thus in addition to the standard items in the Risk of bias tool: selection bias (random sequence generation, allocation concealment); performance bias (blinding of participants and personnel); detection bias (blinding of outcome assessment); attrition bias (incomplete outcome data; consideration of analysis methods, e.g., imputation method); reporting bias (selective reporting); and other sources of bias; We also assessed the following domains as recommended by Moore Size (rating studies with fewer than 50 participants per arm as being at high risk of bias, those with between 50 and 199 participants per arm as being at unclear risk of bias, and 200 or more participants per arm as being at low risk of bias). Duration (rating studies with follow-up of two weeks as being at high risk of bias, two to seven weeks as being at unclear risk of bias and eight weeks or longer as being at low risk of bias). Two review authors completed the Risk of bias assessment for each included study independently. If there was disagreement, the authors met to reach consensus, consulting an independent third review author if necessary. Measures of treatment effect We compared the post-treatment pain intensity scores between the trial arms. Where possible, we calculated the proportion of participants with a specific degree of pain relief and converted it 7

13 into dichotomous information to yield the number of participants who obtained a moderately important benefit (30% pain relief) or a substantially important benefit (50% or more pain relief) as defined by the IMMPACT recommendations (Dworkin 2008). We planned to calculate the risk ratio (RR) as the measure of treatment effect and used this to calculate the number needed to treat for an additional beneficial outcome (NNTB) for 30% and 50% pain relief. We also collected data on the duration of pain relief postintervention where available. For this update, we used the OMERACT 12 group s recommendations for minimally important difference for pain outcomes reported on a continuous scale (Busse 2015). They recommend 10 mm on a mm visual analogue scale (VAS) as the threshold for minimal importance for average between-group change. They highlight that should be interpreted with caution as estimates that fall closely below this point may still reflect a treatment that benefits a considerable number of patients. We used this threshold but interpreted it cautiously. Unit of analysis issues No unit of analysis issues arose since we were unable to conduct a meta-analysis due to insufficient data. Dealing with missing data Where insufficient data were presented to enter a study into the meta-analysis, we contacted study authors to request access to the missing data. Assessment of heterogeneity We planned to assess heterogeneity and its impact using the Chi 2 test and the I 2 test (Higgins 2003; Higgins 2011). Where significant heterogeneity (P < 0.1) was present, we planned to conduct subgroup analyses. Preplanned comparisons included CRPS- I versus CRPS-II, children versus adults, and continuous versus single block. However, no meta-analysis was possible. Assessment of reporting biases We considered the possible influence of publication/small study biases on review findings. For studies that utilised dichotomised outcomes, where possible, we planned to test for the possible influence of publication bias on each outcome by estimating the number of participants in studies with zero effect required to change the NNTB to an unacceptably high level (defined as an NNTB of 10) as outlined by Moore Data synthesis We pooled results where adequate data supported this, using Review Manager 5 software (RevMan 2012). Separate preplanned meta-analyses included sympathetic blockade versus sham/placebo procedure and sympathetic blockade versus no treatment or usual care. We used a random-effects model to combine the studies. We considered separate meta-analyses for shortterm (up to two weeks postintervention), mid-term (more than two to less than seven weeks postintervention) and long-term (seven weeks or longer postintervention) outcomes where we identified adequate data. Assessment of quality of available evidence For this update we used the GRADE approach to assess the quality of evidence (Guyatt 2011a; Guyatt 2011b). Two reviewers independently applied the GRADE criteria to each key comparison. If there was disagreement, the authors met to reach consensus, consulting an independent third review if necessary. We present a summary of our judgements for each comparison in Appendix 5. To ensure consistency of GRADE judgements, we applied the following criteria to each domain equally for all key comparisons of the primary outcome. Limitations of studies: downgrade once if more than 25% of participants were from studies classified as being at a high risk of bias across any domain, excluding the study size domain as this is accounted for in the assessment of imprecision. Inconsistency: downgrade once if heterogeneity is statistically significant and the I 2 value is more than 40%. When a meta-analysis was not performed we downgraded once if trials did not show effects in the same direction. Indirectness: downgrade once if more than 50% of the participants were outside the target group. Imprecision: downgrade once if fewer than 400 participants for continuous data and fewer than 300 events for dichotomous data. Publication bias: downgrade once where there is direct evidence of publication bias or if estimates of effect based on small scale, industry-sponsored studies raising a high index of suspicion of publication bias. Two review authors (NEO, BMW) judged whether these factors were present. We considered single studies to be inconsistent and imprecise, unless more than 400 participants were randomised for continuous outcomes or more than 300 for dichotomous outcomes. We applied the following definitions of the quality of the evidence (Balshem 2011). High quality: We are very confident that the true effect lies close to that of the estimate of the effect. Moderate quality: We are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. 8

14 Low quality: Our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low quality: We have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect. R E S U L T S Description of studies Summary of findings table We included a Summary of findings table to present the main findings in a transparent and simple tabular format. In particular, we included key information concerning the quality of evidence, the magnitude of effect of the interventions examined, and the sum of available data on the outcome pain intensity, hand pain, and duration of pain relief. Subgroup analysis and investigation of heterogeneity We assessed heterogeneity and its impact using the Chi 2 test and the I 2 test (Higgins 2003; Higgins 2011). Where significant heterogeneity (P < 0.1) was present we planned to conduct subgroup analyses. Preplanned comparisons included CRPS-I versus CRPS- II, children versus adults, and repeated versus single blocks. Where possible we used the proportion of people with adverse side effects in each treatment group to calculate the number needed to treat for an additional harmful outcome (NNTH). Sensitivity analysis When sufficient data were available, we conducted sensitivity analyses on the effect of including/excluding studies classified as being at unclear or high risk of bias. Results of the search The previous update of this review included twelve studies (Aydemir 2006; Bonelli 1983; Carroll 2009; Meier 2009; Nascimento 2010; Price 1998, Raja 1991; Rodriguez 2005; Toshniwal 2012; Verdugo 1995, Wehnert 2002; Zeng 2003; combined N = 386). For this update, we included an additional four studies (Freitas 2013; Lim 2007; Rocha 2014; Yoo 2012; combined N = 154]). As our modified criteria excluded studies with follow-up of 48 hours or less, we excluded four studies from this update that had been included in previous versions of this review (Meier 2009; Raja 1991; Verdugo 1995; Wehnert 2002; combined N = 79). Overall, we included 12 studies with 461 participants in this update. One new study is awaiting classification, as it was published as a protocol for a trial and in abstract format only, and it is unclear whether the trial was completed (Kostadinova 2012). Figure 1 presents a flow chart of the search screening process for the present update. We identified 461 studies through the database search strategy and none from searching other sources. After removing duplicates and screening titles and abstracts, we retrieved the full text for five studies. Of these, we included four new studies in the review. 9

15 Figure 1. #Study flow diagram for updated searches 10

16 For this update we attempted to contact the authors of two studies: to retrieve essential data for Freitas 2013 and to check the status of the trial and request a report if available for Kostadinova Included studies We present full details of the studies in the Characteristics of included studies tables. Study participants All included studies evaluated only adult participants (Aydemir 2006; Bonelli 1983; Carroll 2009; Freitas 2013; Lim 2007; Nascimento 2010; Price 1998; Rocha 2014; Rodriguez 2005; Toshniwal 2012; Yoo 2012; Zeng 2003). Nine studies included only people with upper limb CRPS treated with stellate ganglion blockade (Aydemir 2006; Bonelli 1983; Lim 2007; Nascimento 2010; Rocha 2014; Rodriguez 2005; Toshniwal 2012; Yoo 2012; Zeng 2003), and two studies included only people with lower limb CRPS treated with lumbar sympathetic blockade (Carroll 2009; Freitas 2013). The remaining study included a mix of upper and lower limb CRPS (Price 1998). Study designs Two studies used a cross-over design (Carroll 2009; Price 1998), and 10 employed a parallel design (Aydemir 2006; Bonelli 1983; Freitas 2013; Lim 2007; Nascimento 2010; Rocha 2014; Rodriguez 2005; Toshniwal 2012; Yoo 2012; Zeng 2003). All included studies were small, with total numbers of participants ranging from 7 to 82. LASB versus placebo Two studies compared LASB versus placebo (Aydemir 2006; Price 1998). Price 1998 (N = 7) compared stellate ganglion block (n = 4, 15 ml lidocaine 1%) versus lumbar sympathetic block (n = 3, 10 ml bupivacaine 0.125%) with normal saline injection in people with CRPS of the upper or lower extremities based on the IASP diagnostic criteria and investigated the proportion of participants who experienced 50% pain relief. Price 1998 also measured the duration of pain relief and the mean between-group difference in pain relief on a visual analogue scale (VAS). Aydemir 2006 (N = 25) compared stellate ganglion lidocaine block (10 ml lidocaine 1%) plus sham stellate ganglion ultrasound block (n = 9) to a double-sham condition (sham stellate ganglion lidocaine (10 ml saline) and ultrasound blocks). Both groups received rehabilitation treatment. Investigators measured spontaneous pain posttreatment and at one-month follow-up. LASB versus other interventions In contrast to the original version of this review, we included studies, totaling nine, that compared LASB to other interventions (Aydemir 2006; Bonelli 1983; Carroll 2009; Freitas 2013; Lim 2007; Nascimento 2010; Rocha 2014; Toshniwal 2012; Yoo 2012). Aydemir 2006 compared stellate ganglion lidocaine block (10 ml of 1%) plus sham stellate ganglion ultrasound block (n = 9) to stellate ganglion ultrasound block (consisting of ultrasound delivered non-invasively over the stellate ganglion) plus sham stellate ganglion lidocaine block (10 ml of saline; n = 9). Both groups received rehabilitation treatment. Investigators measured the primary outcome of spontaneous pain post-treatment and at onemonth follow-up. Bonelli 1983 (N = 19) compared stellate ganglion block with bupivacaine (15 ml of 0.5%; n = 10) versus intravenous regional blockade (IVRB) with guanethidine (20 mg; n = 9) in patients with reflex sympathetic dystrophy. The primary outcome was the intensity of pain (measured using a 100 mm linear scale) measured post-treatment at 15 minutes, 60 minutes, 24 hours and 48 hours as well as at one and three months. Carroll 2009 (N = 9, of whom seven completed the study) compared sympathetic block with botulinum toxin A (75 units) plus bupivacaine (10 ml of 0.5%) versus bupivacaine alone (10 ml of 0.5%) in people with CRPS of the lower extremity. The primary outcome was the duration that pain (measured using a VAS) remained below baseline levels. Freitas 2013 (N = 40) compared sympathetic block of the lumbar plexus with lidocaine and clonidine versus pulsed radiofrequency treatment of the same structure. Investigators measured pain intensity for up to six months follow-up. Lim 2007 (N = 36) compared a course of five stellate ganglion blocks with lidocaine versus a two-week course of corticosteroids (prednisolone) in patients with CRPS following stroke. They used a self developed four-point scale (0 to 3) of hand pain with passive movement and followed patients up to 30 days from the start of treatment. Nascimento 2010 (N = 43) compared sympathetic block with lidocaine (70 mg 1% lidocaine) versus sympathetic block with lidocaine (70 mg 1% lidocaine) plus clonidine (30 µg) versus IVRB with lidocaine plus clonidine (7.0 ml solution, 1% lidocaine, 1 µg/kg clonidine). Investigators measured intensity of pain (VAS) and duration of pain relief post-treatment and at one-week followup. Rocha 2014 (N = 36) compared image-guided thoracic sympathetic block with ropivacaine and triamcinolone versus injection of the same agents into the subcutaneous space. Authors described this comparison condition as an active control as it might be predicted to induce physiological effects. This allowed blinding of 11

17 participants. Investigators followed up participants using the Brief Pain Inventory as an outcome measure at one month and one year. This study did not report a responder analysis. Toshniwal 2012 compared continuous stellate ganglion block (SGB; n = 18; 280 ml, 0.125% bupivacaine at 2 ml/h for seven days) versus continuous infraclavicular brachial plexus block (n = 12; 400 ml, 0.125% bupivacaine at 5 ml/h for seven days) in people with CRPS-I of the upper extremity. Both groups received concurrent physiotherapy sessions. The primary outcome was the subscale scores on the neuropathic pain scale measured over a fourweek period post-treatment. Yoo 2012 (N = 42) compared stellate ganglion block with image guidance to the same block versus no image guidance in participants with CRPS following stroke. Of note, the group with image guidance received a higher dose of lidocaine (10 ml) than the nonguided group (5 ml). Investigators measured pain intensity with a VAS at two- and four-week follow-up. LASB in addition to other therapies Two studies evaluated the efficacy of LASB as an addition to other therapeutic management (Rodriguez 2005; Zeng 2003). Rodriguez 2005 evaluated physical therapy and pharmacological treatment with or without SGB (N = 41 per group, 10 cc, equal parts 2% lidocaine and 0.5% bupivacaine) in people with upper limb CRPS with a confirmed sympathetic component to their pain (50% pain reduction with screening, prerandomisation SGB). Investigators measured pain intensity, therapeutic efficacy (proportion with at least 50% pain reduction), and relapse rate at two months post-treatment. Zeng 2003 compared SGB (dose not reported) plus rehabilitation versus rehabilitation alone in a group (N = 60) with shoulder-hand syndrome following stroke. Pain (verbal rating scale) was measured at 10 and 20 days posttreatment. Excluded studies In total, we excluded 26 studies. For this update, we excluded one study at the full-text stage as it was not an RCT (Kastler 2013). In the last review we excluded two studies (Rodriguez 2006; Rodriguez 2008), as it was not clear whether they represented original trials in distinct cohorts or an expansion of the included trial by Rodriguez 2005, comprising many of the same participants data. For this update we have reclassified these two studies to Studies awaiting classification and have again attempted to contact the study authors for clarification. See the table Characteristics of excluded studies for details of all studies excluded from all versions of this review. We also identified one further study awaiting classification ( Kostadinova 2012). Risk of bias in included studies We present the summary results of the Risk of bias assessment in Figure 2 and Figure 3. We considered no studies to be at low risk of bias across all domains. 12

18 Figure 2. Risk of bias summary: review authors judgements about each risk of bias item for each included study. 13

19 Figure 3. Risk of bias graph: review authors judgements about each risk of bias item presented as percentages across all included studies. Allocation Only two studies clearly described an adequate randomisation process (Freitas 2013; Toshniwal 2012); we considered the other ten studies to be at unclear risk of bias for this domain. We judged four studies as being at a low risk of bias for allocation concealment (Aydemir 2006; Rocha 2014; Rodriguez 2005; Toshniwal 2012), and we assessed six studies as being at unclear risk of bias (Bonelli 1983; Freitas 2013; Lim 2007; Nascimento 2010; Yoo 2012; Zeng 2003). The remaining studies used a cross-over study design (risk of bias for allocation concealment not applicable). Blinding We considered three studies to have blinded participants and personnel adequately (Aydemir 2006; Carroll 2009; Price 1998) (low risk of performance bias). We considered six studies to be at unclear risk of bias across this domain (Bonelli 1983; Freitas 2013; Nascimento 2010; Rocha 2014; Toshniwal 2012; Yoo 2012) as though the interventions were distinguishable, both were active invasive interventions. Three studies were at high risk of bias (Lim 2007; Rodriguez 2005; Zeng 2003;) as clinicians delivering the interventions were not blinded or the intervention conditions were clearly distinguishable. The outcome of interest for this review was self-reported pain. In this situation, the patient acts as the assessor; therefore risk of detection bias is primarily dependent on participant blinding. For blinding of outcome assessment, we judged five studies to be at low risk of detection bias as they clearly reported blinding of the participants (Aydemir 2006; Carroll 2009; Freitas 2013; Price 1998; Rocha 2014), four studies at unclear risk of bias as it was unclear whether patients were adequately blinded (Bonelli 1983; Nascimento 2010; Toshniwal 2012; Yoo 2012), and three studies were judged to have high risk of detection bias because patients were not adequately blinded (Lim 2007; Rodriguez 2005; Zeng 2003). Incomplete outcome data We considered seven studies to be at unclear risk of bias due to incomplete outcome data (Aydemir 2006; Carroll 2009; Freitas 2013; Lim 2007; Rocha 2014; Rodriguez 2005; Yoo 2012) as a result of the levels of drop-out reported or incomplete reporting of attrition. Selective reporting We judged three studies to be at high risk of bias for this domain due to incomplete reporting of pain scores (Freitas 2013; Price 14

20 1998; Rodriguez 2005). Carroll 2009 carried an unclear risk of bias for incomplete reporting of pain score at a secondary end point. LASB versus placebo For the comparison of LASB versus placebo, we rated all evidence as being of moderate quality. Adequate sample size? We judged all studies to be at high risk of bias with regard to sample size as all had less than 50 participants per arm. Adequate duration of follow-up? We considered all but four studies to be at high or unclear risk of bias based on inadequate duration of follow-up (Bonelli 1983; Freitas 2013; Rocha 2014; Rodriguez 2005). Pain intensity In Price 1998, there was no difference between lidocaine and normal saline; the same number of participants (6/7) achieved at least 50% pain relief at two weeks. In Aydemir 2006, spontaneous pain scores were no different from baseline to post-treatment in either the group receiving lidocaine plus sham ultrasound SGB (Z = 0.18, P = 0.86) or in the group receiving sham lidocaine plus sham ultrasound (Z = 0.76, P = 0.45). Authors did not report between-group comparisons. Other potential sources of bias We judged two studies to be at high risk of bias for other reasons (Bonelli 1983; Rocha 2014). In Bonelli 1983, the LASB group had a significantly shorter duration of symptoms at baseline than the IVRB guanethidine group, and participants were significantly older. Rocha 2014 had average pain scores at baseline that differed by greater than one point between groups, but authors did not present tests for comparability at baseline. Three studies were at unclear risk of bias (Freitas 2013; Yoo 2012). Freitas 2013 and Rodriguez 2005 provided no baseline data, and neither Freitas 2013 nor Yoo 2012 gave details regarding concomitant treatments. We judged the two cross-over studies to be at low risk of bias for carry-over effects (Carroll 2009; Price 1998). There were insufficient data to support a formal statistical analysis of reporting/small study biases for any comparison. Duration of pain relief Price 1998 evaluated the duration of pain relief, finding that when local anaesthetic was administered, the mean duration of relief was longer (three days versus 19.9 hours in the saline group). However, short-term relief was similar in both groups. In Aydemir 2006, spontaneous pain scores were no different from baseline to onemonth follow-up in either the group receiving lidocaine (plus sham ultrasound SGB; Z = 1.05, P = 0.29) or in the group receiving sham lidocaine and sham ultrasound (Z = 0.68, P = 0.50). Authors reported no between-group comparisons. None of the included studies reported postintervention analgesic requirements. Adverse Events Price 1998 and Aydemir 2006 did not report adverse events. Sources of funding and conflicts of interest While not formally included within the Risk of bias assessment, we extracted information regarding study funding and potential conflicts of interest. Seven study reports offered no details regarding these issues (Aydemir 2006; Bonelli 1983; Freitas 2013; Nascimento 2010; Price 1998; Yoo 2012; Zeng 2003). Carroll 2009 declared that the authors had filed a patent for the inclusion of botulinum toxin A in sympathetic blocks. Rodriguez 2005 declared financial support from governmental and nonprofit organisations. No study declared funding from industry sources. Toshniwal 2012 and Rocha 2014 declared no conflict of interest. Effects of interventions See: Summary of findings for the main comparison LASB for pain intensity and duration of pain relief in adults with CRPS For a summary of all core findings, see Summary of findings for the main comparison. LASB versus other interventions Pain relief Most comparative studies reported no significant difference in pain between groups (Bonelli 1983; Freitas 2013; Nascimento 2010; low to very low quality evidence). Aydemir 2006 did not explicitly report between-group differences, although they did not find any within-group differences in spontaneous pain scores between baseline and post-treatment nor at one-month follow-up in either the group receiving lidocaine SGB plus sham ultrasound SGB (Zscores listed above) or in the group receiving ultrasound SGB plus sham lidocaine (Z = 0.59, P = 0.55; Z = 0.63, P = 0.53, respectively; low quality evidence). Due to the variation in the interventions, there were not adequate data to allow pooling of the results. Lim 2007 reported no significant difference in hand pain intensity (scale from 0 to 3) between LASB plus corticosteroid versus oral corticosteroids at 15-day follow-up (mean difference 0.00, 95% 15